A multi-terminal integrated intelligent penetrating supervision analysis and display system and method

Abstract

The application discloses a multi-terminal integrated intelligent penetration type supervision analysis and display system and method, and relates to the technical field of data management and visualization. The system comprises a data aggregation layer, a data management and index modeling layer, a data service and intelligent analysis layer, a multi-terminal differentiated display layer and a user role and permission layer penetrating through all layers, which are connected in turn from bottom to top. The data aggregation layer is responsible for multi-source data access and verification, the data management and index modeling layer outputs index data through hierarchical cleaning standardization and seven-tuple model construction, the data service and intelligent analysis layer provides unified API service and performs intelligent early warning and trend prediction, the multi-terminal differentiated display layer performs differentiated index mapping and adaptive rendering for users with different roles in different terminals, and the user role and permission layer implements full-link permission control. The application can realize data homogeneity, unified caliber and three-terminal collaboration, and improve index traceability, early warning and trend prediction.

Description

Technical Field

[0001] This invention relates to the field of data governance and visualization technology, specifically to a multi-terminal integrated intelligent penetrating monitoring analysis and display system and method. Background Technology

[0002] With the deepening of state-owned asset supervision system reform and the acceleration of enterprise digital transformation, building a comprehensive, end-to-end, and penetrating supervision system has become an inevitable trend in the industry. In current regulatory practices, using digital means to conduct comprehensive and in-depth visualization analysis and display of various business and analytical data is a key path to achieving object penetration, information penetration, and accountability penetration. Existing technologies typically combine the construction of independent data acquisition systems with visualization display platforms, extracting and presenting data scattered across various business systems by deploying terminal devices such as regulatory dashboards and mobile applications. The mainstream technical approach focuses on using big data processing tools to aggregate data in batches and, based on pre-set report templates or basic chart libraries, providing static or near real-time data query and display services to managers at different levels to meet the basic needs of daily regulatory decision-making and business execution.

[0003] However, in the long-term industrial application process, existing technical solutions have gradually exposed some common technical bottlenecks when facing increasingly complex organizational structures and diversified regulatory scenarios. These bottlenecks are mainly reflected in the following aspects: multi-terminal display systems struggle to accommodate the differentiated cognitive needs of different user roles, leading to a disconnect in information transmission between macro-level decision-making and micro-level execution; insufficient depth of integration of multi-source heterogeneous data makes it difficult to support cross-domain correlation analysis; the lack of a unified standardized management mechanism for regulatory indicators results in inconsistent data definitions and difficulty in traceability; and existing systems have limited intelligent analysis capabilities when dealing with massive amounts of data, making it difficult to shift from passive display to proactive early warning and trend prediction. These problems, to some extent, restrict the further improvement of regulatory effectiveness, making the construction of an efficient, collaborative, and intelligent penetrating regulatory closed loop a long-term challenge. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-terminal integrated intelligent penetrating supervision analysis and display system and method to solve the technical problems of multi-terminal display system being unable to take into account the differentiated cognitive needs of different user roles, insufficient depth of multi-source heterogeneous data fusion, lack of unified standardized management mechanism for supervision indicators, and limited intelligent analysis capabilities.

[0005] The implementation process of this invention is as follows: A multi-terminal integrated intelligent penetration-type supervision analysis and display system includes a data aggregation layer module, a data governance and indicator modeling layer module, a data service and intelligent analysis layer module, a multi-terminal differentiated display layer module, and a user role and permission layer module that runs through all the layers, connected sequentially from bottom to top. The data aggregation layer module transmits the collected data to the data governance and indicator modeling layer module; the data governance and indicator modeling layer module transmits the dataset to the data service and intelligent analysis layer module through the data service interface; the data service and intelligent analysis layer module transmits the analysis results to the multi-terminal differentiated display layer module; the user role and permission layer module implements permission control throughout all layers. The data aggregation layer module is used to access the raw data of multiple regulatory business systems, and after performing data quality verification on the raw data, transmit the raw data that has passed the verification to the data governance and indicator modeling layer module. The data governance and indicator modeling layer module is used to perform layered cleaning and standardization on the received raw data to obtain standardized data, then link and integrate it, and construct a seven-tuple standardized regulatory indicator system based on the integrated data. The calculated indicator data is output and a dataset is generated after pre-calculation and summary of the data. The dataset is used by the multi-terminal differentiated display layer module to call and output the dataset to the data service and intelligent analysis layer module. The data service and intelligent analysis layer module is used to provide data services to the outside world in the form of a unified API gateway, and to deploy an intelligent analysis engine to perform early warning signals and predict trends on indicator data. The generated analysis results are output to the multi-terminal differentiated display layer module through API. The multi-terminal differentiated display layer module includes a PC / large screen terminal unit, a PAD terminal unit, and a mobile phone terminal unit. The three types of terminal units correspond to three user roles: leadership decision-making layer, business management layer, and business execution layer, respectively. They are used to call data APIs to obtain datasets and analysis results, and to perform differentiated indicator mapping and adaptive visualization rendering according to the terminal hardware usage scenario and the corresponding user role. The user role and permission layer module is used to perform permission verification at the data access nodes of each layer and control the data access scope and operation permissions according to the user role hierarchy.

[0006] Furthermore, the data aggregation layer module includes a data acquisition unit and a data quality verification unit; the data acquisition unit supports four data access methods: (1) Real-time streaming access: High-frequency updated business data is collected through Kafka message queue, and data latency is controlled within seconds; (2) Batch timed access: collect report-type and summary-type data according to a preset period through the ETL scheduling engine; (3) API interface access: periodically pull data from third-party systems through RESTful API interfaces; (4) File import and access, supporting the parsing of offline reporting data in Excel and CSV formats; The data quality verification unit is used to perform format verification, integrity verification, logical consistency verification, timeliness verification and uniqueness verification during the data access process, and write the data that fails the verification into the abnormal data table for manual review.

[0007] Furthermore, the data governance and indicator modeling layer module includes a data governance unit and an indicator modeling unit; the data governance unit includes a four-layer data architecture, from bottom to top: ODS operational data layer, DWD detailed data layer, DWS summary data layer, and ADS application data layer. The ODS operation data layer is used to receive the raw data output by the data aggregation layer module, maintain the original appearance of the source system data, record the data change history and timestamps, keep the data table structure consistent with the source system, and attach the data source identifier, collection time and data quality verification result to each record, and send the data to the DWD detailed data layer. The DWD detailed data layer is used to perform cleaning processes such as deduplication, missing value filling, and outlier correction on the raw data sent from the ODS operational data layer. Then, the cleaned data is subjected to standardized and unified processing of master data, reference data, and business data to obtain standardized data. Finally, the standardized data is integrated by multi-source data according to the regulatory business theme. The integrated data is input into the seven-tuple standardized regulatory indicator system to perform calculations and obtain the calculated indicator data. The calculated indicator data is then sent to the DWS summary data layer. The indicator modeling unit employs a four-layer decomposition method—strategic objectives, regulatory themes, business domains, and specific indicators—to decompose the overall goal of penetrating supervision into multiple strategic sub-objectives. It then divides these sub-objectives into multiple regulatory business themes, decomposes indicators within each business domain, and finally decomposes quantifiable specific regulatory indicators within each business domain. Based on these specific regulatory indicators, a standardized seven-tuple regulatory indicator system is constructed, and each specific regulatory indicator is defined. The seven-tuple includes indicator ID, statistical caliber, calculation logic, analysis dimension, data source, access level, and refresh strategy. The indicator modeling unit reads the integrated data from the DWD detailed data layer and performs calculations according to the standardized seven-tuple regulatory indicator system to obtain the calculated indicator data. The DWS summary data layer is used to receive the indicator data output by the DWD detailed data layer, summarize it according to the business subject domain, and build a star-shaped dimension model. Each subject contains a fact table and multiple dimension tables. The dimensions include time dimension, organization dimension, industry dimension, and region dimension. The summary data is pre-calculated in the fact table and the dataset is sent to the ADS application data layer. The ADS application data layer receives the aggregated data output from the DWS aggregated data layer and processes it to generate application datasets that can be directly displayed and called by the three terminals: PC / large screen terminal, PAD terminal and mobile terminal. The application datasets include indicator datasets filtered and assembled according to the three-terminal indicator mapping rules, early warning signal datasets generated according to early warning rules, and predicted trend datasets generated according to trend prediction models.

[0008] Furthermore, the data service and intelligent analysis layer module includes a data service unit. This unit provides standardized data services and access to the datasets from the ADS application data layer via an API gateway to three types of terminal units: PC / large screen units, PAD units, and mobile phone units. The APIs include an indicator query API, a multidimensional analysis API, an analysis result API, and a data push API. The indicator query API supports querying indicator values ​​by indicator code, time range, and dimension filtering conditions. The multidimensional analysis API supports OLAP online analysis operations such as drill-down, slicing, and rotation. The analysis result API provides query services for early warning signals and trend prediction results. The data push API proactively pushes data and early warning information to subscribed terminal units when indicator data changes or early warnings are triggered. All three APIs—indicator query API, multidimensional analysis API, analysis result API, and data push API—adopt authentication, rate limiting, and caching mechanisms to ensure consistency in data acquisition across the three terminals. The authentication mechanism uses permission tokens for different user roles within the user role and permission layer module.

[0009] Furthermore, the data service and intelligent analysis layer module also includes an intelligent analysis engine unit, which includes an anomaly detection and early warning submodule, a trend prediction submodule, and a correlation analysis submodule; The anomaly detection and early warning submodule is used to identify indicator anomalies through statistical threshold detection, year-on-year deviation detection and isolated forest algorithm. After classifying the indicator data into first-level, second-level and third-level early warning according to the severity, the warning is pushed to the terminal unit of the corresponding role according to the different levels of warning. The trend prediction submodule is used to perform short-term indicator prediction using the ARIMA time series model for indicator data of 1-3 months, and to perform medium- and long-term indicator prediction using the Prophet model for indicator data of 6-12 months, and output the confidence interval of the prediction results. The association analysis submodule is used to construct an indicator association network of vertical penetration, horizontal linkage, and benchmark comparison of indicator data through association rule mining algorithms. The vertical penetration is the penetration relationship of the same indicator between different organizational levels, the horizontal linkage is the business logic association between different theme indicators, and the benchmark comparison is the comparison relationship between the indicator and the benchmark value.

[0010] Furthermore, the multi-terminal differentiated display layer module executes the following six-dimensional differentiated indicator mapping strategy: In terms of the number of metrics, the PC / large screen unit covers all metric data, the PAD unit covers core metrics, and the mobile phone unit covers key business metrics. In terms of display strategy, the PC / large screen unit adopts a comprehensive presentation and penetrating analysis, the PAD unit adopts a key highlight and layered focus, and the mobile phone unit adopts core filtering and convenient query. In terms of depth of analysis, PC / large screen units support deep drill-down and cross-domain correlation analysis, PAD units support thematic analysis and comparative analysis, and mobile phone units support statistical queries and trend analysis. In terms of interaction methods, the PC / large screen unit supports precise mouse interaction and multi-screen linkage, the PAD unit supports touch interaction and handwriting annotation, and the mobile phone unit supports touch swiping and quick operation. In terms of data density, PC / large screen units are displayed at high density, PAD units at medium density, and mobile phone units at low density. In terms of refresh strategy, PC / large screen units use real-time refresh at the second level, PAD units use timed refresh at the minute level, and mobile phone units use push refresh on demand.

[0011] Furthermore, the PC / large screen unit in the multi-terminal differentiated display layer module is geared towards the leadership decision-making level for command and dispatch and overall decision-making. Its core functions are macro-level regulatory data statistical analysis, global situational awareness, and cross-domain correlation analysis. The PAD unit is geared towards the business management level for daily supervision, meeting reports, and special viewing. Its core functions are viewing data of the business segments under its jurisdiction, special analysis, and meeting presentations. The mobile terminal unit is geared towards the business execution level for daily office work, mobile queries, and instant processing. Its core functions are business data statistical queries, early warning message processing, and task supervision.

[0012] Furthermore, the multi-terminal differentiated display layer module executes an adaptive visualization rendering mechanism, specifically: The layout is adaptive, with PC / large screen units using a grid-based free layout, PAD units using a horizontal column layout, and mobile phone units using a vertical flow layout. The charts are adaptive. The PC / large screen unit supports visual charts such as heat maps, Sankey diagrams, relationship diagrams and 3D maps. The PAD unit supports interactive charts. The mobile unit supports line charts, bar charts and donut charts. The interaction is adaptive. The PC / large screen unit supports mouse hover for details, right-click menu, and drag-and-drop drill-down interaction. The PAD unit supports touch click and handwritten annotation interaction. The mobile phone unit supports touch click, long press for details, two-finger zoom, and pull-to-refresh interaction.

[0013] Furthermore, the hierarchical permission control rules for the user roles and permission layer modules are as follows: the leadership decision-making layer corresponds to the PC / large screen terminal unit, which has access to all regulatory indicators and global decision-making operation permissions; the business management layer corresponds to the PAD terminal unit, which has access to thematic indicators of the business domain under its jurisdiction and the operation permissions for thematic analysis and meeting reporting; the business execution layer corresponds to the mobile terminal unit, which has access to relevant business indicators and the operation permissions for data query and task processing; permission verification is performed in the API gateway of the data service, and the scope of indicators and operation permissions visible to the corresponding user are filtered based on the user role token.

[0014] The method for multi-terminal integrated intelligent analysis and display using the above system includes the following steps: S1, Multi-source data aggregation It accesses raw data from multiple regulatory business systems, performs data quality verification on the raw data, and outputs qualified data that passes the verification. S2, Hierarchical Data Governance and Indicator Modeling The qualified raw data is subjected to hierarchical cleaning, standardization and correlation integration. Based on the integrated data, a seven-tuple standardized regulatory indicator system is constructed, and the calculated indicator data is generated. The indicator data is pre-calculated and summarized, and the summarized data is used to generate a dataset. The dataset is used for the multi-terminal differentiated display layer module and output to the data service and intelligent analysis layer module. S3, Unified Data Services and Intelligent Analytics The dataset is provided to the outside world as a standardized data service in the form of a unified API gateway. At the same time, the intelligent analysis engine performs early warning signal and predictive trend analysis on the indicator data, and the generated analysis results are output to the outside world through API. S4. Multi-terminal differentiated display: PC / large screen terminal, PAD terminal and mobile terminal respectively call data API to obtain datasets and analysis results. The three types of terminals correspond to three user roles: leadership decision-making level, business management level and business execution level. Differentiated indicator mapping and adaptive visualization rendering are performed according to the terminal hardware characteristics and corresponding user roles. S5. Full-link hierarchical access control: Performs access control verification at every node in the data access chain, and controls the scope of data access and operation permissions according to user roles.

[0015] The beneficial effects of this invention are: (1) By constructing a data aggregation layer module, a data governance and indicator modeling layer module, a data service and intelligent analysis layer module, a multi-terminal differentiated display layer module, and a user role and permission layer module that runs through each layer, a complete data flow and control closed loop is formed, which effectively solves the problem that the multi-terminal display system is difficult to take into account the differentiated cognitive needs of different user roles. (2) The original data of multiple regulatory business systems are accessed through the data aggregation layer module and strict quality verification is performed to ensure the accuracy and integrity of the input data. Then, the data governance and indicator modeling layer module performs layered cleaning and standardization processing on the original data. By constructing a seven-tuple standardized regulatory indicator system, the unified modeling and calculation of indicators are realized, thereby outputting standardized data assets and indicator data, which solves the problems of insufficient fusion depth of multi-source heterogeneous data and lack of a unified standardized management mechanism for regulatory indicators. (3) Data services are provided to the outside world through the data service and intelligent analysis layer module using a unified API gateway, and an intelligent analysis engine is deployed to generate early warning signals and perform trend prediction analysis on indicator data. The in-depth analysis results are output to the multi-terminal differentiated display layer module in real time. With the help of the user role and permission layer module, permission verification is performed at each data access node. This avoids the delay in regulatory decision-making caused by inconsistent data standards, homogeneous display content and single analysis methods. It improves the consistency of regulatory data and the accuracy of indicator traceability, and enhances the overall efficiency and response speed of penetrating supervision. Attached Figure Description

[0016] Figure 1 It is a multi-terminal integrated intelligent penetration-based monitoring, analysis, and display system; Figure 2 This is a schematic diagram of the data aggregation layer module and the data governance unit; Figure 3 Example diagram of indicator modeling unit; Figure 4 Schematic diagram of the data service and intelligent analysis layer module; Figure 5 A diagram illustrating the differences in user roles and functions across the three platforms; Figure 6 This is a schematic diagram illustrating the differentiating dimensions of the multi-terminal differentiated display layer module. Detailed Implementation

[0017] The present invention will be further described below with reference to the embodiments.

[0018] Example 1 A multi-terminal integrated intelligent penetration-based monitoring analysis and display system, see Figure 1-6It includes a data aggregation layer module, a data governance and indicator modeling layer module, a data service and intelligent analysis layer module, a multi-terminal differentiated display layer module, and a user role and permission layer module that runs through the settings of each layer module, which are connected sequentially from bottom to top. The data aggregation layer module transmits the collected data to the data governance and indicator modeling layer module; the data governance and indicator modeling layer module transmits the dataset to the data service and intelligent analysis layer module through the data service interface; the data service and intelligent analysis layer module transmits the analysis results to the multi-terminal differentiated display layer module; the user role and permission layer module implements permission control throughout all layers. (i) The data aggregation layer module is used to access raw data from various regulatory business systems. After performing data quality verification on the raw data, the verified raw data is transmitted to the data governance and indicator modeling layer module. The data aggregation layer module serves as a data cleaning pool and buffer. It first accesses the raw data from various regulatory business systems through a configurable adapter. After access, the module's built-in data quality verification engine scans the raw data. Verification rules may include null value detection, value range checks, and logical correlation verification. Only raw data that passes verification is allowed to be transmitted to the data governance and indicator modeling layer module for subsequent processing. Data that fails verification is marked and isolated for administrators to investigate or correct anomalies, thereby ensuring the purity of data assets entering the core governance process.

[0019] The data aggregation layer module includes a data acquisition unit and a data quality verification unit; the data acquisition unit supports four data access methods: (1) Real-time streaming access: High-frequency updated business data is collected through Kafka message queue, and the data latency is controlled at the second level. Specifically, Kafka consumer service is deployed in the data middle platform. High-frequency updated business data (such as large-amount fund transaction data, real-time financial data, etc.) are collected in streaming through Kafka message queue, and the collected data is written to the ODS operation data layer in real time.

[0020] (2) Batch timed access: collect report-type and summary-type data according to a preset period through the ETL scheduling engine; specifically, batch collection of report-type and summary-type data (such as monthly financial statements, quarterly investment summaries, etc.) is achieved through the DataX / SqoopETL scheduling engine.

[0021] (3) API interface access: Data from third-party systems is retrieved periodically through RESTful API interfaces. Data from third-party systems (such as external credit data, industry statistics data, etc.) is retrieved periodically through standard RESTful APIs.

[0022] (4) File import access, supports parsing offline reporting data in Excel and CSV formats; for offline reporting data (such as manually filled Excel reports, etc.), it supports importing files in Excel, CSV and other formats through the file parser.

[0023] The data quality verification unit is used to perform format verification, integrity verification, logical consistency verification, timeliness verification and uniqueness verification during the data access process, and write the data that fails the verification into the abnormal data table for manual review.

[0024] The specific verification rules include: ① Format validation: Validate whether the data field type, length, and precision conform to the preset specifications (e.g., the amount field must be a numeric type with a precision of 2 decimal places, and the date field must be in YYYY-MM-DD format). ② Integrity check: Check whether the required fields are empty, and key fields (such as enterprise code and indicator code) must not be missing; ③ Logical consistency check: Check whether the logical relationship between the data is reasonable (e.g., total profit should equal operating profit plus net non-operating income and expenses, and the debt-to-equity ratio should be between 0% and 100%). ④ Timeliness verification: Verify whether the data update time is within the valid time window (e.g., monthly report data should be submitted within the 5th of the following month); ⑤ Uniqueness check: Check whether data records with the same business primary key are duplicated (the same indicator value of the same enterprise in the same month should exist uniquely).

[0025] (ii) The data governance and indicator modeling layer module is used to perform layered cleaning and standardization on the received raw data to obtain standardized data, then link and integrate it, and construct a seven-tuple standardized regulatory indicator system based on the integrated data. The calculated indicator data is output and a dataset is generated after pre-calculation and summary of the data. The dataset is used by the multi-terminal differentiated display layer module to call and output the dataset to the data service and intelligent analysis layer module. The data governance and indicator modeling layer module includes a data governance unit and an indicator modeling unit. The data governance unit comprises a four-layer data architecture, from bottom to top: ODS operational data layer, DWD detailed data layer, DWS summary data layer, and ADS application data layer. The ODS operation data layer is used to receive the raw data output by the data aggregation layer module, maintain the original appearance of the source system data, record the data change history and timestamps, keep the data table structure consistent with the source system, and attach the data source identifier, collection time and data quality verification result to each record, and send the data to the DWD detailed data layer. The DWD detailed data layer is used to perform cleaning processes such as deduplication, missing value filling, and outlier correction on the raw data sent from the ODS operational data layer. Then, the cleaned data is subjected to standardized and unified processing of master data, reference data, and business data to obtain standardized data. Finally, the standardized data is integrated by multi-source data according to the regulatory business theme. The integrated data is input into the seven-tuple standardized regulatory indicator system to perform calculations and obtain the calculated indicator data. The calculated indicator data is then sent to the DWS summary data layer. The cleaning process employs commonly used data cleaning methods in the prior art, which will not be discussed in detail here.

[0026] The standardization and unification process includes: (1) Master data standardization: Establish an enterprise master data table, assign a unified enterprise code to each regulated enterprise (the coding rule is "administrative division code + industry code + enterprise serial number"), map different enterprise names and different enterprise codes in various business systems to a unified code, and solve the problem of inconsistent codes of the same enterprise in different systems; establish an organization personnel master data table to unify personnel codes and organizational hierarchy mapping.

[0027] (2) Reference data standardization: Establish industry classification standard code table (unified adoption of national economic industry classification GB / T4754 standard), regional coding standard table (unified adoption of administrative division code GB / T2260 standard), measurement unit standard table, etc., and map different classification standards in various business systems to unified standard codes.

[0028] (3) Business data standardization: Unify data format (dates are unified in YYYY-MM-DD format, amounts are unified in RMB and the precision is 2 decimal places), unify statistical standards (clarify the statistical standards of each indicator, such as whether "operating revenue" includes non-operating income, whether it offsets internal transactions, etc.).

[0029] Association and integration involves associating and piecing together standardized data from different business domains according to business themes or entity objects to form data assets from a holistic perspective. Specifically, it involves associating and integrating cleaned and standardized data according to business themes to establish an enterprise fact table (containing basic enterprise information, financial data, investment data, and other related information), providing a foundation for thematic modeling of the DWS aggregated data layer.

[0030] The indicator modeling unit employs a four-layer decomposition method—strategic objectives, regulatory themes, business domains, and specific indicators—to break down the overall goal of penetrating supervision into multiple strategic sub-objectives. Multiple regulatory business themes are then defined around each strategic sub-objective, and indicators are decomposed by business domain within each theme. Finally, quantifiable specific regulatory indicators are decomposed within each business domain. A standardized seven-tuple regulatory indicator system is constructed based on these specific indicators, and each specific regulatory indicator is defined. The seven-tuple includes indicator ID, statistical caliber, calculation logic, analysis dimension, data source, access control level, and refresh strategy. The indicator modeling unit reads the integrated data from the DWD detailed data layer and performs calculations according to the standardized seven-tuple regulatory indicator system to obtain the calculated indicator data. This model-based management eliminates indicator ambiguity and ensures the uniqueness and consistency of indicators throughout the system. Finally, this module outputs the indicator data calculated based on the seven-tuple model. The standardized seven-tuple regulatory indicator system here is a system construction method in existing technologies; the difference lies in replacing the elements of the existing seven-tuples with the indicator ID, statistical caliber, calculation logic, analysis dimension, data source, access control level, and refresh strategy of this invention. The construction process involved in the system is the same as that of existing technologies, and will not be described in detail.

[0031] The seven-tuple is specifically: —Indicator ID: Assign a globally unique identifier code to each indicator. The coding rule is "subject code-domain code-indicator serial number", such as "FIN-PROFIT-001" which represents the ROE indicator under the financial subject profitability domain. —Statistical scope: Clearly define the calculation range of the indicators, the inclusion / exclusion conditions, and the data value rules; —Calculation logic: Defines the calculation formula and steps for the indicators, supporting both formulaic expression and programmatic implementation; —Analysis Dimensions: Define the analysis dimensions supported by this indicator, including time dimension (year / quarter / month / day), organizational dimension (group / segment / enterprise), industry dimension, and geographic dimension, etc.; —Data source: Clearly identify the source system and data table of the indicator data, and indicate the data access method and update frequency; —Access control hierarchy: Defines the data security level of this metric and the access permissions for different user roles; —Refresh strategy: Define the refresh frequency of this indicator on the three platforms. Real-time indicators are pushed every second, regular indicators are refreshed on a T+1 timer, and historical indicators are loaded on demand.

[0032] like Figure 3The decomposition method is illustrated below: a four-tiered decomposition approach is adopted, consisting of "strategic objectives → regulatory themes → business domains → specific indicators." First, the overall goal of penetrating supervision is decomposed into three strategic sub-goals: economic operation, reform advancement, and innovative development. Then, 14 regulatory business themes are defined around these three strategic sub-goals (including state-owned asset overview, financial supervision, investment supervision, debt supervision, property rights operation, large-scale fund monitoring, major decision-making, board governance, enterprise reform, and technological innovation). Next, indicators are decomposed under each regulatory business theme according to business domains (e.g., financial supervision is decomposed into profitability, solvency, operational efficiency, and development capabilities). Finally, specific regulatory indicators are derived under each business domain. This ultimately forms a three-tiered indicator system covering over 700 specific regulatory indicators across the 14 regulatory business themes.

[0033] The DWS summary data layer is used to receive the indicator data output by the DWD detailed data layer, summarize it according to the business subject domain, and build a star-shaped dimension model. Each subject contains a fact table and multiple dimension tables. The dimensions include time dimension, organization dimension, industry dimension, and region dimension. The summary data is pre-calculated in the fact table and the dataset is sent to the ADS application data layer. The ADS application data layer receives the aggregated data output from the DWS aggregated data layer and processes it to generate application datasets that can be directly displayed and called by the three terminals: PC / large screen terminal, PAD terminal and mobile terminal. The application datasets include indicator datasets filtered and assembled according to the three terminal indicator mapping rules, early warning signal datasets generated according to early warning rules, and predicted trend datasets generated according to trend prediction models. The data governance and indicator modeling layer module transmits standardized data and indicator data to the data service and intelligent analysis layer module through a data service interface. In this structure, the data governance and indicator modeling layer module interacts with the upper layer through the standardized data service interface, achieving decoupling between data production and data consumption. This interface-based transmission method shields the underlying storage details, provides a unified data access standard, and facilitates flexible invocation by upper-layer applications. The specific implementation of the data service interface can be a RESTful API, Web Service, etc., and this embodiment does not impose any special limitations on it.

[0034] (iii) The data service and intelligent analysis layer module is used to provide data services to the outside world in the form of a unified API gateway, and to deploy intelligent analysis engine units to perform early warning signals and predict trends on indicator data. The generated analysis results are output to the multi-terminal differentiated display layer module through API. The data service and intelligent analysis layer module includes a data service unit. This unit provides standardized data services and access to the datasets from the ADS application data layer to three types of terminal units: PC / large screen units, PAD units, and mobile phone units, via an API gateway. The APIs include an indicator query API, a multidimensional analysis API, an analysis result API, and a data push API. The indicator query API supports querying indicator values ​​by indicator code, time range, and dimension filtering conditions. The multidimensional analysis API supports OLAP online analysis operations such as dimension drill-down, slicing, and rotation. The analysis result API provides query services for early warning signals and trend prediction results. The data push API proactively pushes data and early warning information to subscribed terminal units when indicator data changes or early warnings are triggered. All three APIs—indicator query API, multidimensional analysis API, analysis result API, and data push API—adopt authentication, rate limiting, and caching mechanisms to ensure consistency in data acquisition across the three terminals. The authentication mechanism uses permission tokens for different user roles within the user role and permission layer module.

[0035] The data service and intelligent analysis layer module further includes an intelligent analysis engine unit, which includes an anomaly detection and early warning submodule, a trend prediction submodule, and a correlation analysis submodule. The anomaly detection and early warning submodule is used to identify indicator anomalies through statistical threshold detection, year-on-year deviation detection and isolated forest algorithm. After classifying the indicator data into first-level, second-level and third-level early warning according to the severity, the warning is pushed to the terminal unit of the corresponding role according to the different levels of warning. The specific implementation method is as follows: Statistical threshold detection: Based on historical data, a statistical baseline value (mean μ and standard deviation σ) is calculated for each indicator. When the current period value of an indicator deviates from the baseline value by more than a preset threshold (i.e., the current period value exceeds the range of [μ-3σ, μ+3σ]), it is marked as an outlier. For example, if a company's monthly operating revenue has a mean of 50 million yuan and a standard deviation of 3 million yuan, then a monthly operating revenue lower than 41 million yuan or higher than 59 million yuan is marked as an anomaly. Statistical threshold detection is suitable for anomaly determination of single indicators.

[0036] Year-on-Year Deviation Detection: This function calculates the deviation of the current period's value from the historical value of the same period (year-on-year deviation rate). An alert is triggered when the deviation exceeds a preset percentage threshold. For example, if a company's debt-to-asset ratio increases by more than 10% year-on-year this quarter, a deviation alert is triggered. Year-on-year deviation detection is suitable for identifying abnormal trend deviations in indicators.

[0037] The Isolation Forest algorithm for comprehensive anomaly detection (using a known algorithm from existing technology, which will not be detailed here): First, multiple relevant regulatory indicators (such as debt-to-asset ratio, revenue growth rate, cash flow ratio, investment completion rate, etc.) are used as feature dimensions to construct feature vectors. Second, the feature vector space is randomly and recursively partitioned multiple times using the Isolation Forest model. Each time, a feature dimension is randomly selected, and a partition point is randomly selected within the range of values ​​for that dimension, dividing the data space into multiple sub-regions. Then, the "anomaly score" of the feature vector is calculated. The anomaly score is determined by the average path length required for the feature vector to be partitioned and isolated in the Isolation Forest. The shorter the path (i.e., the easier it is to be partitioned and isolated), the more abnormal the data point is. Finally, indicators with anomaly scores exceeding a preset threshold are marked as comprehensive anomaly objects. The advantage of the Isolation Forest algorithm is that it can consider the combined anomaly features of multiple indicators simultaneously, making it suitable for scenarios where "a single indicator seems normal, but the combination of multiple indicators is abnormal."

[0038] The warning results are divided into three levels according to severity: Level 1 Warning (Attention Level) – the indicator deviates from the normal range but does not reach the threshold; Level 2 Warning (Warning Level) – the indicator reaches the preset threshold or is abnormal for multiple consecutive periods; Level 3 Warning (Severe Level) – the indicator exceeds the red line or triggers the compliance bottom line. Warnings of different levels are pushed to users on the three platforms with different roles according to preset routes.

[0039] The trend prediction submodule is used to perform short-term indicator prediction using the ARIMA time series model for indicator data of 1-3 months, and to perform medium- and long-term indicator prediction using the Prophet model for indicator data of 6-12 months, and output the confidence interval of the prediction results. Specifically, for short-term forecasts (1-3 months), an ARIMA (Autoregressive Differential Moving Average) time series model is used, with rolling forecasts based on recent historical data. First, historical time series data of the indicator is collected (e.g., the operating revenue data of a company for the past 24 months), and the stationarity of the time series data is tested (e.g., ADF test). If the data is not stationary, differencing is performed. Second, the order parameters (p and q) of AR and MA are determined through autocorrelation and partial autocorrelation plots, with the differencing order being d, forming an ARIMA(p,d,q) model (an existing model, used here for application only). Then, historical data is used to estimate and fit the model parameters, and residual tests are performed to confirm the model's effectiveness. Finally, based on the fitted model, rolling forecasts of the indicator values ​​for the next 1-3 months are made, with the actual values ​​updated monthly, the model refitted, and the forecast results updated.

[0040] For medium- to long-term forecasts (6-12 months), the Prophet model is used, comprehensively considering trend, seasonality, and holiday effects. First, long-term historical data of the indicators (e.g., monthly data from the past 3-5 years) is collected. Second, model parameters are configured—specifying the detection sensitivity of trend change points, the seasonality cycle parameter (annual cycle of 365.25 days), and a custom list of holidays (e.g., special dates that may affect state-owned asset supervision data, such as Spring Festival and National Day). Then, the model automatically identifies trend turning points and seasonal patterns in the historical data, fitting trend functions, seasonal functions, and holiday effect functions. Finally, the three components are superimposed to generate forecast values ​​for the next 6-12 months, while simultaneously outputting the confidence intervals (upper and lower limits) to assist decision-makers in assessing the uncertainty of the forecast results.

[0041] The predicted results are continuously compared with the actual values ​​to analyze the deviation, and the model parameters are automatically adjusted to improve the prediction accuracy.

[0042] The association analysis submodule is used to construct an indicator association network of vertical penetration, horizontal linkage, and benchmark comparison of indicator data through association rule mining algorithms. The vertical penetration is the penetration relationship of the same indicator between different organizational levels, the horizontal linkage is the business logic association between different theme indicators, and the benchmark comparison is the comparison relationship between the indicator and the benchmark value.

[0043] The specific implementation method is as follows: using association rule mining algorithms, high-frequency association patterns (such as association rules such as "high debt-to-asset ratio and low cash flow frequently appearing at the same time") are discovered in a large number of business records, and an association network between indicators is established, including vertical association (the penetrating relationship of the same indicator between different organizational levels), horizontal association (the business logic relationship between different thematic indicators), and benchmarking association (the comparison relationship between indicators and benchmark values), supporting multi-dimensional association analysis and root cause tracing.

[0044] (iv) The multi-terminal differentiated display layer module includes a PC / large screen terminal unit, a PAD terminal unit and a mobile phone terminal unit. The three types of terminal units correspond to three user roles: leadership decision-making layer, business management layer and business execution layer, respectively. They are used to call data API to obtain indicator data and analysis results, and to perform differentiated indicator mapping and adaptive visualization rendering according to the terminal hardware usage scenario and the corresponding user role. The multi-terminal differentiated display layer module implements the following six-dimensional differentiated indicator mapping strategy: In terms of the number of indicators, the PC / large screen unit covers all indicator data, the PAD unit covers core indicators, and the mobile unit covers key business indicators. Because the PC / large screen unit has a large display space and high-performance hardware support, it can handle massive amounts of information, thus covering all indicator data. For example, it can display over 700 indicators across all regulatory themes to meet the needs of leadership decision-makers for a comprehensive understanding of the overall situation. The PAD unit is geared towards business management, focusing primarily on the daily supervision of their respective business segments. Therefore, it covers core indicators, such as selecting over 200 key indicators closely related to management decisions to highlight key areas. The mobile unit is geared towards the business execution layer, mainly used for on-site verification and immediate processing. Limited by screen size and mobile scenarios, it covers key business indicators, such as displaying only over 300 frequently queried key indicators to ensure information simplicity and efficient access.

[0045] In terms of presentation strategy, PC / large screen units adopt a comprehensive presentation and penetrating analysis approach, PAD units emphasize key points and focus on specific levels, and mobile units employ core filtering and convenient querying. The PC / large screen units utilize a comprehensive presentation strategy, showcasing the entire regulatory panorama on one or more screens through macro-level aggregation, global overview, multi-dimensional comparison, and trend analysis, supporting penetrating analysis—that is, drilling down from the macro-level overview to detailed data. PAD units employ a emphasizing key points and focusing on specific levels, conducting thematic analysis on specific regulatory topics, displaying in-depth information through drill-down linkage and comparative ranking, while also incorporating warning lists to assist decision-making. Mobile units employ a core filtering and convenient querying strategy, displaying key information in the form of numerical overviews, trend charts, and red, yellow, and green lights, and providing warning cards to ensure users can quickly access core information in mobile scenarios.

[0046] In terms of analytical depth, the PC / large screen units support deep drill-down and cross-domain correlation analysis, the PAD units support thematic analysis and comparative analysis, and the mobile units support statistical queries and trend analysis. The PC / large screen units support strategic-level deep drill-down and cross-domain correlation analysis, enabling global analysis and cross-topic analysis to support macro-level decision-making. The PAD units support management-level thematic analysis and comparative analysis, allowing drilling down to third-level enterprises for in-depth thematic analysis, meeting the business oversight needs of middle managers. The mobile units support execution-level statistical queries and trend analysis, primarily used for detailed queries targeting specific business processes and simple trend tracking.

[0047] In terms of interaction methods, the PC / large screen unit supports precise mouse interaction and multi-screen linkage, the PAD unit supports touch interaction and handwritten annotation, and the mobile phone unit supports touch swiping and quick operations. The PC / large screen unit supports large-screen touch, remote control operation, and precise mouse interaction, enabling full-screen display, carousel, and multi-screen linkage, suitable for large-scale operations in command centers. The PAD unit supports touch screen interaction, including gesture zoom, swipe switching, and drill-down, and specifically supports handwritten annotation, facilitating marking and explanation in meeting presentations. The mobile phone unit supports swipe browsing, click viewing, pull-to-refresh, and event-triggered push operations, conforming to the habits of one-handed operation and rapid response on mobile devices.

[0048] In terms of data density, PC / large screen units are displayed at high density, PAD units at medium density, and mobile units at low density. PC / large screen units feature high density, accommodating over 20 chart components per screen, suitable for presenting complex data relationships on ultra-high resolution screens. PAD units use medium density, displaying 4 to 6 charts per page, maintaining adequate white space to balance information volume and reading comfort. Mobile units feature low density, displaying only 1 to 3 cards per screen, using large fonts and high contrast to ensure clear readability in outdoor or mobile environments.

[0049] In terms of refresh strategy, PC / large screen units use real-time refresh at the second level, PAD units use timed refresh at the minute level, and mobile units use push refresh on demand. PC / large screen units use real-time refresh at the second level or timed batch refresh (e.g., T+1, hourly) to ensure the command center can obtain the latest situational information. PAD units use near real-time refresh, such as refreshing every 15 minutes or on demand, adapting to the pace of office scenarios. Mobile units use real-time push refresh, i.e., refresh at the second level or push immediately when a specific event (such as an alert trigger) occurs, ensuring the business execution layer can receive key information in a timely manner.

[0050] The multi-terminal differentiated display layer module includes a PC / large screen unit for leadership decision-making, used for command and dispatch and overall decision-making. Its core functions are macro-level regulatory data statistical analysis, global situational awareness, and cross-domain correlation analysis, displaying over 700 indicators across 14 regulatory themes. The PAD unit is for business management, used for daily supervision, meeting reports, and special reviews. Its core functions are viewing data for the business segments under its jurisdiction, thematic analysis, and meeting presentations, displaying over 200 core indicators across 9 key themes. The mobile terminal unit is for business execution, used for daily office work, mobile queries, and instant processing. Its core functions are business data statistical queries, early warning message processing, and task supervision, displaying over 300 key indicators across various business themes.

[0051] The multi-terminal differentiated display layer module executes an adaptive visualization rendering mechanism, specifically... The layout is adaptive. PC / large-screen units utilize a grid-based free layout to support high-resolution multi-component display. PAD units employ a horizontal column layout to suit 10-12 inch landscape screen usage habits, while mobile phone units use a vertical flow layout to suit 5-6.5 inch portrait screen usage habits. For PC / large-screen units, leveraging their high resolution and large display area, the grid-based free layout supports parallel display and free combination of multiple components, facilitating global situational awareness and macro-level comparison of cross-domain data. For PAD units, adapting to their medium screen size and landscape usage habits, the horizontal column layout can rationally divide information hierarchy, enabling focused display of key content. For mobile phone units, addressing the limitations of their small portrait screen size, the vertical flow layout ensures content is arranged logically from top to bottom, conforming to the operating habits of one-handed swiping and improving the efficiency of obtaining key information.

[0052] The charts are adaptive. PC / large-screen units support heatmaps, Sankey diagrams, relationship diagrams, and 3D maps; PAD units support interactive charts; and mobile units support line charts, bar charts, and donut charts. Leveraging powerful graphics processing capabilities, PC / large-screen units render heatmaps to present data distribution density, use Sankey diagrams to show flow and proportion, reveal complex network relationships through relationship diagrams, and utilize 3D maps to present a macro-level geographic spatial dimension, meeting the needs of in-depth analysis. PAD units support interactive charts of moderate complexity, balancing visual richness with touch responsiveness, and supporting filtering and interactive elements. Mobile units prioritize lightweight charts such as line charts, bar charts, and donut charts. These charts have clear structures and can quickly convey core information such as trends, comparisons, and proportions within limited screen space, avoiding information overload. The interactive interface is adaptive. The PC / large-screen unit supports mouse hover for details, right-click menus, and drag-and-drop drill-down. The PAD unit supports touch clicks and handwritten annotations. The mobile unit supports touch clicks, long-press for details, two-finger zoom, and pull-to-refresh. The PC / large-screen unit, based on high-precision mouse pointer positioning, supports hovering to view details, right-clicking to access extended options, and dragging for data drill-down, suitable for refined data exploration. The PAD unit combines touchscreen and stylus input, offering handwritten annotation functionality in addition to basic touch clicks, facilitating marking and recording during meeting presentations or business reviews. The mobile unit, based on finger touch characteristics, provides a convenient and smooth mobile operating experience through long-press to access details, two-finger zoom to view local details, and pull-to-refresh to get the latest data.

[0053] (v) The user role and permission layer module is used to perform permission verification at the data access nodes of each layer and control the data access scope and operation permissions according to the user role hierarchy.

[0054] The hierarchical permission control rules for the user roles and permission layer modules are as follows: The leadership decision-making layer corresponds to the PC / large screen unit, which can view all indicators (over 700 items) across all topics, possessing access to all regulatory indicators and global decision-making operation permissions; the business management layer corresponds to the PAD unit, which can view thematic indicators within its business domain (e.g., the financial supervision department can only view financial thematic indicators, and the investment supervision department can only see investment thematic indicators), possessing access to thematic indicators within its business domain and the operation permissions for thematic analysis and meeting reporting; the business execution layer corresponds to the mobile terminal unit, which can view business indicators related to its job duties, possessing access to relevant business indicators and the operation permissions for data query and task processing. Permission verification is performed in the data service's API gateway, filtering the visible indicator range and operation permissions for the corresponding user based on user role tokens. Based on user roles and permission hierarchy, permission verification is implemented in the data service module's API gateway. Each API request must carry a user role token. The API gateway filters the visible indicator range for that user based on the role information in the token, ensuring that users with different roles can only obtain indicator data and analysis results within their permission range.

[0055] This permission verification mechanism is the execution means to ensure the implementation of the aforementioned hierarchical permission control rules, deployed at the unified API gateway of the data service and intelligent analysis layer modules. When a PC / large screen unit, PAD unit, or mobile phone unit initiates a data request, it must include a valid user role token in the request header. This token is issued by the system during the user login phase according to the user's role (i.e., leadership decision-making level, business management level, or business execution level), and encapsulates identity and role attribute information. After receiving the request, the API gateway first verifies the legality and validity of the token and parses the user's role level. Subsequently, the API gateway dynamically generates the set of indicators visible to the user and the list of allowed operation permissions based on the pre-configured hierarchical permission control rules. For example, for a token parsed as belonging to the business management level, the API gateway will automatically filter out indicator data outside the relevant business domain, returning only thematic indicators within its permission scope, and blocking global decision-making operation interfaces. Through this token-based dynamic filtering mechanism, the system can precisely control the boundaries of each data interaction, effectively preventing unauthorized access and data leakage, thereby ensuring the compliance and security of transparent monitoring data.

[0056] Example 2: Construction of a Multi-Terminal Integrated Intelligent Analysis and Display System for State-Owned Asset Supervision in a Certain Province Taking the State-owned Assets Supervision and Administration Commission of a certain province's construction of a penetrating supervision three-terminal display and analysis system as an example, the implementation process of this invention is explained in detail.

[0057] Step 1: Building the Data Aggregation Layer First, a data platform infrastructure was built, deploying a distributed data acquisition engine to access data from all of the organization's existing and newly established regulatory business systems, including the Major Decision-Making System, Project Investment Management System, Financial Supervision System, Property Rights Management System, Technological Innovation System, Policy and Regulations System, Board of Directors Management System, and Supervision and Accountability System. Real-time data was streamed via Kafka message queues, batch data was updated periodically via DataX / Sqoop, third-party data was retrieved periodically via RESTful APIs, and offline data was automatically parsed via file import interfaces. All collected data was written to the ODS layer after quality verification; data that failed verification was written to an exception data table for manual review.

[0058] Step 2: Data Governance and Indicator Modeling The raw ODS data is managed according to a four-layer architecture: ODS→DWD→DWS→ADS. —The ODS layer retains the original data and records the data change history and timestamps.

[0059] —The DWD layer performs cleaning (duplicate removal, missing value filling, outlier correction), standardization (establishing unified coding for enterprise master data, unified mapping for industry classifications, and unified data format specifications), and correlation integration (establishing enterprise fact tables according to business themes). Simultaneously, an indicator system is designed using a four-layer decomposition method: the overall goal of penetrating supervision is decomposed into three strategic sub-goals, further divided into 14 regulatory business themes, each theme is decomposed by business domain, ultimately defining over 700 specific indicators. Each indicator is standardized and modeled using a seven-tuple standardized regulatory indicator system. The integrated data is processed through a standardized regulatory indicator system of seven tuples to obtain the calculated indicator data, which is then sent to the DWS aggregated data layer.

[0060] —The DWS layer constructs a star-shaped dimensional model based on 14 regulatory themes and pre-computes aggregated data.

[0061] —The ADS layer displays the application dataset (indicator dataset, early warning signal dataset, prediction trend dataset, etc.) generated from the requirements to three terminals.

[0062] Step 3: Deployment of Data Services and Intelligent Analytics Deploy the data service module, providing four core services: indicator query API, multidimensional analysis API, analysis result API, and data push API. Deploy the intelligent analysis engine, loading three analysis sub-modules: —Anomaly detection and early warning: Anomaly indicators are detected based on statistical thresholds and the isolated forest algorithm, and the warnings are pushed to the corresponding three users according to the three-level warning system.

[0063] —Trend Forecasting: Short-term and medium-to-long-term forecasts of key indicators based on ARIMA and Prophet models.

[0064] —Correlation analysis: Discover the correlation rules between indicators and establish a correlation network that allows for vertical penetration, horizontal linkage, and comparative analysis.

[0065] Step 4: Configure differentiated display across three platforms —PC / Large Screen Configuration: Configure a comprehensive overview large screen (including total state-owned assets, structural distribution, early warning summary, etc.) and 14 regulatory theme large screens (each theme is configured with 20-40 visualization components, including trend charts, distribution charts, ranking charts, heat maps, etc.), configure a drill-down path (comprehensive overview → domain monitoring → enterprise profile → indicator penetration), and verify more than 700 indicators in total.

[0066] —PAD configuration: Configure special display pages for 9 key themes (Overview of State-owned Assets, Economic Operation, Capital Layout, Central-Local Cooperation, Industrial Development, State-owned Enterprise Reform, Technological Innovation, and Party Building), and verify more than 200 key indicators.

[0067] —Mobile configuration: Configure data statistics query pages for 14 business themes, configure mobile-exclusive functions such as early warning push message cards, to-do task list, quick operation entry, and verify more than 300 core business indicators.

[0068] Step 5: Configure User Roles and Permissions Configure three user tiers for access control: leadership decision-makers can view all metrics, business management personnel can view metrics for their respective business domains, and business execution personnel can view metrics relevant to their specific roles. Implement permission verification in the API gateway to ensure that different roles can only access data within their authorized scope.

[0069] Example 3: Dynamic Optimization of Regulatory Data Indicator System Taking the optimization of the indicator system of a provincial state-owned assets supervision system after one year of operation as an example, this paper illustrates the life cycle management and dynamic optimization method of the indicator system.

[0070] After one year of system operation, an evaluation of the indicator system revealed that: more than 50 indicators had extremely low usage frequency (less than 5 visits per quarter), 35 indicators had a data quality score of less than 70 points, and 25 new regulatory requirements had not yet been included in the indicator system.

[0071] Accordingly, the indicator system was optimized as follows: 30 low-value indicators were removed and archived; 35 indicators with substandard data quality were revised (data sources were optimized or calculation methods were adjusted); 25 new indicators were added (covering new policy requirements and new regulatory business areas); and the permission levels of 15 indicators were adjusted (permission configuration was optimized based on actual usage feedback). The optimized indicator system was released as a new version under the version management of the indicator management center, and the indicator configuration was automatically updated across the three systems without requiring code changes.

[0072] Example 4: Application of Typical Regulatory Scenarios Taking the scenario of "monitoring abnormal large-scale fund transactions of a provincial state-owned enterprise" as an example, this paper illustrates the application process of the three-terminal system of the present invention in specific regulatory business.

[0073] Scenario Description: The regulatory system needs to monitor large-scale fund transactions of provincial state-owned enterprises in real time, identify abnormal transaction behaviors, and handle them.

[0074] Application process: (1) Large-amount abnormal transaction related indicators (about 20 items) are written to the ODS layer in real time through the Kafka of the data aggregation layer; (2) After being cleaned and standardized by the data governance layer, they are provided to the three terminals through the data service API; (3) The abnormal detection sub-module of the intelligent analysis engine detects abnormal transactions in real time and automatically triggers warnings; (4) The PC terminal displays the transaction flow map and abnormal transaction heat distribution in real time; (5) The PAD terminal leaders can view the specific enterprise transaction details and issue verification instructions; (6) The mobile terminal business personnel receive the verification task, go to the site for verification and report the results; (7) The large screen terminal and PAD terminal update the handling status to form a complete regulatory closed loop.

[0075] Implementation results: The time to detect abnormal transactions has been reduced from days to minutes; the average duration of verification tasks has been reduced from 72 hours to 8 hours; and the completeness rate of regulatory closed loop has reached over 98%.

[0076] Example 5 The method for multi-terminal integrated intelligent analysis and display using the system described in Example 1 includes the following steps: S1, Multi-source data aggregation It accesses raw data from multiple regulatory business systems, performs data quality verification on the raw data, and outputs qualified data that passes the verification. S2, Hierarchical Data Governance and Indicator Modeling The qualified raw data is subjected to hierarchical cleaning, standardization and correlation integration. Based on the integrated data, a seven-tuple standardized regulatory indicator system is constructed, and the calculated indicator data is generated. The indicator data is pre-calculated and summarized, and the summarized data is used to generate a dataset. The dataset is used for the multi-terminal differentiated display layer module and output to the data service and intelligent analysis layer module. S3, Unified Data Services and Intelligent Analytics The dataset is provided to the outside world as a standardized data service in the form of a unified API gateway. At the same time, the intelligent analysis engine performs early warning signal and predictive trend analysis on the indicator data, and the generated analysis results are output to the outside world through API. S4. Multi-terminal differentiated display: PC / large screen terminal, PAD terminal and mobile terminal respectively call data API to obtain datasets and analysis results. The three types of terminals correspond to three user roles: leadership decision-making level, business management level and business execution level. Differentiated indicator mapping and adaptive visualization rendering are performed according to the terminal hardware characteristics and corresponding user roles. S5. Full-link hierarchical access control: Performs access control verification at every node in the data access chain, and controls the scope of data access and operation permissions according to user roles.

[0077] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A multi-terminal integrated intelligent penetrating monitoring, analysis, and display system, characterized in that: It includes a data aggregation layer module, a data governance and indicator modeling layer module, a data service and intelligent analysis layer module, a multi-terminal differentiated display layer module, and a user role and permission layer module that runs through the settings of each layer module, which are connected from bottom to top. The data aggregation layer module transmits the collected data to the data governance and indicator modeling layer module; The data governance and indicator modeling layer module transmits the dataset to the data service and intelligent analysis layer module through the data service interface; the data service and intelligent analysis layer module transmits the analysis results to the multi-terminal differentiated display layer module. The user role and permission layer module implements permission control across all layers; The data aggregation layer module is used to access the raw data of multiple regulatory business systems, and after performing data quality verification on the raw data, transmit the raw data that has passed the verification to the data governance and indicator modeling layer module. The data governance and indicator modeling layer module is used to perform layered cleaning and standardization on the received raw data to obtain standardized data, then link and integrate it, and construct a seven-tuple standardized regulatory indicator system based on the integrated data. The calculated indicator data is output and a dataset is generated after pre-calculation and summary of the data. The dataset is used by the multi-terminal differentiated display layer module to call and output the dataset to the data service and intelligent analysis layer module. The data service and intelligent analysis layer module is used to provide data services to the outside world in the form of a unified API gateway, and to deploy an intelligent analysis engine to perform early warning signals and predict trends on indicator data. The generated analysis results are output to the multi-terminal differentiated display layer module through API. The multi-terminal differentiated display layer module includes a PC / large screen terminal unit, a PAD terminal unit, and a mobile phone terminal unit. The three types of terminal units correspond to three user roles: leadership decision-making layer, business management layer, and business execution layer, respectively. They are used to call data APIs to obtain datasets and analysis results, and to perform differentiated indicator mapping and adaptive visualization rendering according to the terminal hardware usage scenario and the corresponding user role. The user role and permission layer module is used to perform permission verification at the data access nodes of each layer and control the data access scope and operation permissions according to the user role hierarchy.

2. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The data aggregation layer module includes a data acquisition unit and a data quality verification unit; the data acquisition unit supports four data access methods: (1) Real-time streaming access: High-frequency updated business data is collected through Kafka message queue, and data latency is controlled within seconds; (2) Batch timed access: collect report-type and summary-type data according to a preset period through the ETL scheduling engine; (3) API interface access: periodically pull data from third-party systems through RESTful API interfaces; (4) File import and access, supporting the parsing of offline reporting data in Excel and CSV formats; The data quality verification unit is used to perform format verification, integrity verification, logical consistency verification, timeliness verification and uniqueness verification during the data access process, and write the data that fails the verification into the abnormal data table for manual review.

3. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The data governance and indicator modeling layer module includes a data governance unit and an indicator modeling unit; the data governance unit includes a four-layer data architecture, from bottom to top: ODS operational data layer, DWD detailed data layer, DWS summary data layer, and ADS application data layer. The ODS operation data layer is used to receive the raw data output by the data aggregation layer module, maintain the original appearance of the source system data, record the data change history and timestamps, keep the data table structure consistent with the source system, and attach the data source identifier, collection time and data quality verification result to each record, and send the data to the DWD detailed data layer. The DWD detailed data layer is used to perform cleaning processes such as deduplication, missing value filling, and outlier correction on the raw data sent from the ODS operational data layer. Then, the cleaned data is subjected to standardized and unified processing of master data, reference data, and business data to obtain standardized data. Finally, the standardized data is integrated by multi-source data according to the regulatory business theme. The integrated data is input into the seven-tuple standardized regulatory indicator system to perform calculations and obtain the calculated indicator data. The calculated indicator data is then sent to the DWS summary data layer. The indicator modeling unit employs a four-layer decomposition method—strategic objectives, regulatory themes, business domains, and specific indicators—to decompose the overall goal of penetrating supervision into multiple strategic sub-objectives. It then divides these sub-objectives into multiple regulatory business themes, decomposes indicators within each business domain, and finally decomposes quantifiable specific regulatory indicators within each business domain. Based on these specific regulatory indicators, a standardized seven-tuple regulatory indicator system is constructed, and each specific regulatory indicator is defined. The seven-tuple includes indicator ID, statistical caliber, calculation logic, analysis dimension, data source, access level, and refresh strategy. The indicator modeling unit reads the integrated data from the DWD detailed data layer and performs calculations according to the standardized seven-tuple regulatory indicator system to obtain the calculated indicator data. The DWS summary data layer is used to receive the indicator data output by the DWD detailed data layer, summarize it according to the business subject domain, and build a star-shaped dimension model. Each subject contains a fact table and multiple dimension tables. The dimensions include time dimension, organization dimension, industry dimension, and region dimension. The summary data is pre-calculated in the fact table and then sent to the ADS application data layer. The ADS application data layer receives the aggregated data output from the DWS aggregated data layer and processes it to generate application datasets that can be directly displayed and called by the three terminals: PC / large screen terminal, PAD terminal and mobile terminal. The application datasets include indicator datasets filtered and assembled according to the three-terminal indicator mapping rules, early warning signal datasets generated according to early warning rules, and predicted trend datasets generated according to trend prediction models.

4. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The data service and intelligent analysis layer module includes a data service unit, which provides standardized data services and access to the datasets of the ADS application data layer to three types of terminal units: PC / large screen terminal, PAD terminal, and mobile terminal, in the form of an API gateway; the API includes indicator query API, multidimensional analysis API, analysis result API, and data push API. The indicator query API supports querying indicator values ​​by indicator code, time range, and dimension filtering conditions; the multidimensional analysis API supports OLAP online analysis operations such as drill-down, slicing, and rotation. The analysis results API is used to provide query services for early warning signals and trend prediction results; The data push API is used to proactively push data and warning information to subscribed terminal units when indicator data changes or warnings are triggered. The indicator query API, multidimensional analysis API, analysis result API, and data push API all adopt authentication, rate limiting, and caching mechanisms to ensure the consistency of data obtained by the three terminals. The authentication is the permission token of different user roles in the user role and permission layer module.

5. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 4, characterized in that: The data service and intelligent analysis layer module also includes an intelligent analysis engine unit, which includes an anomaly detection and early warning submodule, a trend prediction submodule, and a correlation analysis submodule. The anomaly detection and early warning submodule is used to identify indicator anomalies through statistical threshold detection, year-on-year deviation detection and isolated forest algorithm. After classifying the indicator data into first-level, second-level and third-level early warning according to the severity, the warning is pushed to the terminal unit of the corresponding role according to the different levels of warning. The trend prediction submodule is used to perform short-term indicator prediction using the ARIMA time series model for indicator data of 1-3 months, and to perform medium- and long-term indicator prediction using the Prophet model for indicator data of 6-12 months, and output the confidence interval of the prediction results. The association analysis submodule is used to construct an indicator association network of vertical penetration, horizontal linkage, and benchmark comparison of indicator data through association rule mining algorithms. The vertical penetration is the penetration relationship of the same indicator between different organizational levels, the horizontal linkage is the business logic association between different theme indicators, and the benchmark comparison is the comparison relationship between the indicator and the benchmark value.

6. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The multi-terminal differentiated display layer module executes the following six-dimensional differentiated indicator mapping strategy: In terms of the number of metrics, the PC / large screen unit covers all metric data, the PAD unit covers core metrics, and the mobile phone unit covers key business metrics. In terms of display strategy, the PC / large screen unit adopts a comprehensive presentation and penetrating analysis, the PAD unit adopts a key highlight and layered focus, and the mobile phone unit adopts core filtering and convenient query. In terms of depth of analysis, PC / large screen units support deep drill-down and cross-domain correlation analysis, PAD units support thematic analysis and comparative analysis, and mobile phone units support statistical queries and trend analysis. In terms of interaction methods, the PC / large screen unit supports precise mouse interaction and multi-screen linkage, the PAD unit supports touch interaction and handwriting annotation, and the mobile phone unit supports touch swiping and quick operation. In terms of data density, PC / large screen units are displayed at high density, PAD units at medium density, and mobile phone units at low density. In terms of refresh strategy, PC / large screen units use real-time refresh at the second level, PAD units use timed refresh at the minute level, and mobile phone units use push refresh on demand.

7. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The multi-terminal differentiated display layer module includes a PC / large screen unit for leadership decision-making, used for command and dispatch and overall decision-making, with core functions including macro-level regulatory data statistical analysis, global situational awareness, and cross-domain correlation analysis; a PAD unit for business management, used for daily supervision, meeting reports, and special viewing, with core functions including viewing data of the business segments under their supervision, special analysis, and meeting presentations; and a mobile terminal unit for business execution, used for daily office work, mobile queries, and instant processing, with core functions including business data statistical queries, early warning message processing, and task supervision.

8. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The multi-terminal differentiated display layer module executes an adaptive visualization rendering mechanism, specifically... The layout is adaptive, with PC / large screen units using a grid-based free layout, PAD units using a horizontal column layout, and mobile phone units using a vertical flow layout. The charts are adaptive. The PC / large screen unit supports visual charts such as heat maps, Sankey diagrams, relationship diagrams and 3D maps. The PAD unit supports interactive charts. The mobile unit supports line charts, bar charts and donut charts. The interaction is adaptive. The PC / large screen unit supports mouse hover for details, right-click menu, and drag-and-drop drill-down interaction. The PAD unit supports touch click and handwritten annotation interaction. The mobile phone unit supports touch click, long press for details, two-finger zoom, and pull-to-refresh interaction.

9. The multi-terminal integrated intelligent penetrating monitoring analysis and display system according to claim 1, characterized in that: The hierarchical permission control rules for the user roles and permission layer modules are as follows: the leadership decision-making layer corresponds to the PC / large screen terminal unit, which has access to all regulatory indicators and global decision-making operation permissions; the business management layer corresponds to the PAD terminal unit, which has access to the special indicators of the business domain under its jurisdiction and operation permissions for special analysis and meeting reporting; and the business execution layer corresponds to the mobile terminal unit, which has access to the relevant business indicators of its own job and operation permissions for data query and task processing. Permission verification is performed in the API gateway of the data service, filtering the scope of metrics visible to the corresponding user and operation permissions based on the user role token.

10. A method for multi-terminal integrated intelligent analysis and display using the system described in claim 1, characterized in that, Includes the following steps: S1, Multi-source data aggregation It accesses raw data from multiple regulatory business systems, performs data quality verification on the raw data, and outputs qualified data that passes the verification. S2, Hierarchical Data Governance and Indicator Modeling The qualified raw data is subjected to hierarchical cleaning, standardization and correlation integration. Based on the integrated data, a seven-tuple standardized regulatory indicator system is constructed, and the calculated indicator data is generated. The indicator data is pre-calculated and summarized, and the summarized data is used to generate a dataset. The dataset is used for the multi-terminal differentiated display layer module and output to the data service and intelligent analysis layer module. S3, Unified Data Services and Intelligent Analytics The dataset is provided to the outside world as a standardized data service in the form of a unified API gateway. At the same time, the intelligent analysis engine performs early warning signal and predictive trend analysis on the indicator data, and the generated analysis results are output to the outside world through API. S4. Multi-terminal differentiated display: PC / large screen terminal, PAD terminal and mobile terminal respectively call data API to obtain datasets and analysis results. The three types of terminals correspond to three user roles: leadership decision-making level, business management level and business execution level. Differentiated indicator mapping and adaptive visualization rendering are performed according to the terminal hardware characteristics and corresponding user roles. S5. Full-link hierarchical access control: Performs access control verification at every node in the data access chain, and controls the scope of data access and operation permissions according to user roles.

Images